Video-based fall detection plays a crucial role in telemonitoring as a key component in ensuring timely intervention and safety for older adults who live alone. Despite promising advances, most vision-based fall detection methods remain insufficiently robust and fail to generalize effectively to real-world scenarios. In this paper, we present a comprehensive experimental framework designed to rigorously and generically assess the robustness of fall detection approaches. This framework encompasses three evaluation settings: intra-subject, intersubject, and cross-dataset evaluation. It is validated using a custom architecture that combines a convolutional neural network (CNN) with a bidirectional LSTM (BiLSTM), evaluated on three public datasets: URF, Le2i-FD, and MCFD. Experimental results demonstrate strong generalization, with recall exceeding $80 \%$ in cross-dataset settings. These findings underscore the importance of diverse evaluation strategies in developing reliable fall detection systems.

Hospital bed requirement prediction is essential for efficient healthcare resource management and delivering high-quality patient care. Unscheduled admissions significantly impact waiting times and complicate bed management strategies. In this paper, we present different a Multi-Task Learning (MTL) models to predict bed requirements for various Inpatient Departments (IPD) simultaneously. Specifically, we target prediction of daily bed requirements for patients admitted through the Emergency Department (ED). The proposed MTL approach, demonstrates significant advantages in predictive accuracy. Among implemented models, the MTL with XGBoost outperformed than other models in 11 tasks out of 12 with sMAPE, MAE and MSE values ranging from 0.185 to 0.974 and MTL with PyTorch model achieved low sMAPE value of 0.620 with 1 task. A key contribution of this paper is our simultaneous predictive approach, that considers the interaction between departments and leverages the bed occupancy data from the previous day (lag 1) to make a prediction. This methodology aims to reduce waiting time, optimize resource utilization and enhance patient flow within the hospital by following unscheduled admissions. This study demonstrates the potential of AI-driven predictive analytics in hospital resource management and emphasize the significance of MTL methodologies in predicting bed requirements.

Optical Character Recognition (OCR) is extremely useful in various sectors for exploring massive archived data. This technology enables the digitization of printed and handwritten texts that are frequently present in the medical field. For instance, medication error (ME) reports were previously and still in some healthcare facilities written manually, this has led to the accumulation of numerous handwritten data that are unfortunately challenging to exploit. Their digitization through OCR allows extracting important data from these documents and using them to populate the database to implement future analysis techniques to optimize the medication error management process. This paper presents a transformer-based handwritten recognition architecture that employs the Transformer-Based Optical Character Recognition (TrOCR) model combined with image segmentation techniques. Although the TrOCR model provided by Microsoft performs reasonably well in handwritten recognition, it is limited to English text because its pretrained version was trained exclusively on English samples. This limitation is problematic for us, as our task involves digitizing French medication dictation errors. Additionally, its limitation to processing single-line text images impairs its ability to recognize paragraphs. To address these limitations, we will fine-tune the model on French handwritten data and integrate a single-line level segmentation technique, thereby overcoming these constraints. Therefore, the preliminary results from implementing our proposed architecture are promising for the digitization of medication error reports.

Human Activity Recognition (HAR) is essential for applications like health monitoring and fall detection using mobile sensors. However, obtaining large datasets necessary for training HAR systems is prohibitively expensive. Generative Adversarial Networks (GANs) have been proposed to generate synthetic HAR data, simplifying and enhancing the development of HAR systems. While these methods have improved accuracy, existing GAN-based approaches struggle with generating clear abnormal patterns, weakening anomaly detection capabilities. Available HAR data predominantly focuses on normal activities and does not target anomalies, making it challenging to detect anomalous situations. To address this, we propose a novel controllable GAN that generates realistic HAR data as well as distinct abnormal classes. This advancement enhances anomaly detection accuracy through more standardized activity recognition and quantification by HAR systems. We evaluate our approach on the WISDM dataset, demonstrating significant improvements in the balance and quality of synthetic data, leading to better performance in anomaly detection.

The growing elderly population in developed countries highlights the critical need for preventing frailty, which poses significant challenges to health systems due to increased risks of severe health issues. This paper introduces Senselife, a framework that provides explainable service recommendations specifically tailored for frailty prevention. It begins by outlining the medical context and challenges associated with aging, followed by an overview of existing recommender systems with similar objectives. We detail the integration of three key resources-ROR, RNA, and Data Laregion-within the Senselife framework to represent service supply. The paper explains how service supply is structured and the transformation of available data for use within our recommender engine. We introduce the concept of operational activities derived from the ROR and leverage the capabilities of LLMs to incorporate RNA data into Senselife. Additionally, we illustrate how these services are ultimately compiled into recommended service packages. Finally, the paper concludes by summarizing key findings and suggesting potential directions for future research.

Efficient patient transport is crucial for the smooth operation of hospitals, as it directly impacts the timely delivery of medical care and the overall patient experience. The complexity of managing transport logistics, especially with limited resources, necessitates advanced optimization techniques. In this paper, we present an optimization model designed to enhance the scheduling of patient transportation missions and improve the organization of transport activities. Our primary objective is to minimize delays and reduce the number of unaccomplished missions, even when transporter availability is constrained. Using historical data from a hospital in France and the CPLEX solver, we compare our model's performance with existing models in the literature. The results demonstrate that our model significantly improves efficiency and reliability, ensuring timely patient transport and better resource utilization.

Because of their potentially serious consequences, Medication errors (ME) represent a major challenge for health-care facilities. To manage these errors and minimize their seriousness, healthcare professionals follow a collaborative management process that relies on reporting and then analyzing the reports filled in by them via various reporting tools, both digital and paper-based. These tools can be customized requiring specific information to be entered in multiple forms with commonly the presence of textual descriptions to fill in a free-text field. Therefore, text analysis is crucial for thoroughly understanding and effectively analyzing medication errors. Given the large volume of reports to be quickly processed, it is essential to help healthcare professionals prioritize which ME to analyze. In this context, we propose, in this work, processing ME reports with natural language processing tasks using Transformer models such as GPT. In this study, we present the extraction of key information from the reports to help structure textual descriptions of ME with the GPT-4 transformer model. The results obtained show the potential of this model to extract relevant information from ME descriptions in French language without any deep fine-tuning,

The Medication Use Process (MUP) plays a vital role in ensuring the appropriate and safe use of medications. However, this process is exposed to many risks in particular medication errors. These later present a significant challenge, that may have negative consequences for patients. Implementing a medication error reporting and analyzing tool would facilitate an in-depth analysis of the causes of errors and enable the implementation of corrective measures. Additionally, a specific risk management model for medication management, which includes regular monitoring and preventive actions, is essential to prevent medication errors. By implementing these measures, We guarantee the attainment of secure and efficient management of MUP, resulting in better therapeutic results and heightened patient safety. In this work, we propose a tool that helps healthcare professionals create reports and analyze medication errors. This tool has been made and tested with healthcare professionals of the Intercommunal Hospital Center Castres-Mazamet (CHIC). This collaborative effort aimed not only to detect areas for improvement but also to instigate positive change in patient care. By leveraging our innovative tool, healthcare providers were empowered to proactively analyze and report medication errors.

Frailty presents a significant global health challenge for older adults, necessitating effective interventions to support healthy aging. Technology-based solutions, particularly recommender systems, hold promise in addressing frailty prevention. This paper presents Senselife platform for service recommendation dedicated for frailty prevention.We delve into the challenges associated with developing an engaging and user-friendly recommender system specifically tailored for frailty prevention. Key challenges we address include implementing effective data collection strategies and designing user-centered interfaces. Our proposed recommendation platform leverages self-evaluation to deliver personalized recommendations, with the goal of enhancing the functional capabilities of older adults. By aligning the available services within the elderly environment with the demands they face, our solution tackles the complexities of managing frailty in this population.Throughout this study, we elucidate the construction of our surveys, the main source of data for Senselife and the design considerations behind our user interfaces, highlighting our efforts in overcoming the unique challenges associated with systems dedicated to elderly usage.

Medication errors associated with the Medication Use Process present significant risks and require effective management. However, current practices do not address these risks. There is a lack of awareness that medication errors are unintended risks, and a lack of dedicated systems to manage them. To overcome these limitations, a digital system exploiting massive medical data (big data) is proposed. By integrating various data sources, this system aims to provide adaptable medication errors’ management and continuous improvement. This article presents an overview of the system requirements and highlights the potential of Data Mining in healthcare. Implementing this system could revolutionize medication error management and improve patient safety.

Abstract Frailty is a clinical syndrome associated with ageing that characterizes an intermediate state between robust health and loss of autonomy. To preserve the abilities of older adults and prevent dependency, it is important to identify and evaluate their frailty. This approach is part of a dependency prevention strategy, based on a thorough understanding of their medical, social, and living environment. This understanding is usually acquired through significant data collection using standardized evaluation surveys. The obtained data is then analyzed to provide personalized recommendations for the beneficiaries’ lifestyles. Our article presents the concept of frailty and a personalized recommendation system aimed at helping citizens prevent frailty. This system uses an innovative self-assessment approach designed for older adults, without necessarily involving healthcare professionals.

Finding the exact gene mutations that cause a genetic disease has been a challenging task. Despite the development in information technology, the task of extracting gene-disease associations has been mainly a manual process. This is a time-consuming process, in which experts extract gene-disease associations from relevant research papers from the literature manually. The main aim of this paper is to develop an automated approach for extracting and classifying gene-disease associations from relevant literature research papers using both natural language processing and machine learning techniques. This paper extracted data from free-text literature research papers and built four different dataset formats to discover an optimal representation. Machine and Deep learning models (NB, KNN, SVM, NN, CNN, and LSTM) with TF-IDF were applied on the built datasets. As a result, the format of the dataset with (Positive and Negative) instances only, was found to be the best representation for extracting gene-disease associations with optimal accuracy between 74% and 91%. For the four dataset representations, Multilayer Neural Networks was able to predict all classes in most experiments with accuracy between 64% and 91%. From the initial results, this work highlights the need for additional work to improve both the performance of these models and the data extraction method to build more accurate and optimal dataset representation.

A virtual mirrored replica of the real-world, has become a new trend in recent years with the advent of Industry 4.0, shows how intelligent digital twins (DT) can unlock busi-ness value. The implementation of DT requires various types of technologies, including the Internet of Things (IoT), cloud computing, artificial intelligence, and others. Today, DTs can be used in various fields, including manufacturing, smart cities and healthcare. It can be used to monitor the real environment, predict its future, control its behavior, and improve its overall performance. Despite the aforementioned advantages of DTs, if not protected, they can also be considered as an open environment for attackers. If attackers take over a DT, they may end up owning the real environment controlled by the DT. This can result in damaging consequences. This paper explores the potential security and privacy threats that may be brought in through DTs. It proposes and presents a threat meta-model of DTs that identifies key security aspects of concern.

Improving patient safety and quality of care has become a priority for healthcare organizations, given the frequency and potential severe clinical consequences of medication errors and adverse drug events during the Medication-use Process. Medication error reporting can be one most effective strategies to achieve these goals. Indeed, high error reporting rates indicate a positive safety culture rather than an unsafe healthcare environment. It is through the identification of these errors safety barriers can be put in place to prevent a similar event from occurring in the future. However, current healthcare organizations still suffer from a lack of attention in this context, especially in establishing a digital tool dedicated to medication error reporting, which is the motivation for the work described in this paper. The latter proposes a novel tool to make the local reporting of medication errors easier and encourage reporting these errors with the healthcare professionals' confidentially. The overall response to the tool provided was positive from the staff participating.

In this study, we address the performance evaluation of a multi-model call center. We provide an in-depth statistical data analysis to understand the dynamics of waiting times, service times and arrivals. Afterwards, we present our methodology on how to better size and schedule the agents in order to maintain a better service quality, namely more efficient utilization of resources and shorter waiting times.

There have been numerous attempts to build query generators that compute eligibility criteria (EC) for a clinical trial automatically on repositories of patient data. However, one of the challenging key features of EC is the ability to express and compute complex temporal aspects. Existing EC generators has limited temporal capability and those do rely on underlying database technology to perform temporal reasoning. We propose a model that incorporates temporal features of existing generators. However, it separates the computation of the criteria, and in particular the temporal semantics, from the extraction of clinical data from the database to increase the efficiency of execution. We explain the implementation of this model and in particular its temporal algorithm, which runs in O(n log(n)) time where n is the number of clinical facts stored making it more efficient than existing reported generators, where performance, at best, has been reported to be O(n2). We perform an empirical validation to demonstrate the results.

Healthcare organizations are environments of high management complexity and are subject to risk. Indeed, risk management is one of the most relevant aspects put forward in the literature which highlights the necessity to perform comprehensive analyses intended to uncover the root causes of risks. However, the healthcare sector still suffers from a lack of attention in this context, especially with regard to the establishment of risk management and process-oriented management, which is the motivation for the study described in this paper. In light of these observations, it would be essential for healthcare organizations to explore new risk management approaches. Contributing to this field, the present paper applies a risk-aware business process management method to work out a systemic methodology to study risks impacting healthcare processes. This framework aims to improve healthcare organizations’ maturity towards risk management. A case study related to the management of potential risks in a given healthcare process shall illustrate the usage of the developed framework.

Abstract Risk-aware Business Process Management (R-BPM) has been addressed in research since more than a decade. However, the integration of the two independent research streams is still ongoing with a lack of research focusing on the conceptual modeling perspective. Such an integration results in an increased meta-model complexity and a higher entry barrier for modelers in creating conceptual models and for addressees of the models in comprehending them. Multi-view modeling can reduce this complexity by providing multiple interdependent viewpoints that, all together, represent a complex system. Each viewpoint only covers those concepts that are necessary to separate the different concerns of stakeholders. However, adopting multi-view modeling discloses a number of challenges particularly related to managing consistency which is threatened by semantic and syntactic overlaps between the viewpoints. Moreover, usability and efficiency of multi-view modeling have never been systematically evaluated. This paper reports on the conceptualization, implementation, and empirical evaluation of e-BPRIM, a multi-view modeling extension of the Business Process-Risk Management-Integrated Method (BPRIM). The findings of our research contribute to theory by showing, that multi-view modeling outperforms diagram-oriented modeling by means of usability and efficiency of modeling, and quality of models. Moreover, the developed modeling tool is openly available, allowing its adoption and use in R-BPM practice. Eventually, the detailed presentation of the conceptualization serves as a blueprint for other researchers aiming to harness multi-view modeling.

Today, a majority of the elderly want to live longer in autonomy and comfort. Since there is not enough space available in specialized institutions, Home Health Care Services (HHCS) constitute an important addition that reinforces the traditional system. With the increase in HHCS demands, the main challenge in this distributed system is the organization of care services in an HHC institution. This organizational problem is widely studied as a Home Health Care Scheduling and Routing Problem (HHCSRP) in the Operations Research (OR) community. Facing the diversity and complexity of demands, as done in OR, the formulated solution-oriented model aims at providing decision making support for the decision maker in HHC systems. This leads obstacles in the dataset / benchmark based experimental environment and critical analysis such as the degree of the optimality regarding generated solutions and the comprehensive visualization of solutions. In this paper, based on a systemic analysis of the existing benchmark and the solution approach for HHCSRP, a prototype named BLP AD. Core is introduced. It provides a full decision support system to the decision-maker in the HHC system. The functionality consists of an instance generation based on the real dataset, operational planning generation and solution visualization. This research highlights the scientific challenges and attempts to respond to the complicated expectations of HHC systems in a comprehensive manner.

Frailty is a clinical syndrome relative to ageing, and characterizes an intermediate state between robust health and the loss of autonomy. The preservation of one's capacities or abilities falls within a dependency prevention approach, based on a thorough understanding of one's medical and social situation, and his/her environment of life. This understanding is usually gained through an extensive collection of data, using standardized assessment surveys. The obtained data is next analyzed in order to provide personalized recommendations relative to the ways of living. In this paper, we describe the concept, architecture and use of an ICT platform which aims to support professionals and patients in the assessment of frailty at home. Our prototype consists of a knowledge-based system relative to assessment surveys which is connected to a smartphone application and IoT devices.

Nowadays, the majority of elderly people want to live longer in autonomy and well-being. As there are not enough places available in specialized institutions, Home Health Care Services are an alternative. With the fast rise of the demand, the organizations become aware of both existing limitations and of this potential for Research and Development (R&D). They manifest two types of need: one is to be able to identify their own HHC coordination problem, the other is to find solutions for this specific need. In this paper, we propose a systemic analysis of these two aspects to identify the complexity of the relationship between the offer and the demand for HHC services. Then, we present one prototype BLPAD for decision-making support of the coordination in HHC. This research work reveals scientific obstacles and attempts to define a comprehensive response to this expectation of the HHC services ecosystem.

To offer high quality services to patients, hospitals try to get a perception of patients' processes. This research work aims at identifying a type of processes known as patients' pathways and to devise a framework which could support the task of discovering patients' pathways. Existing modeling languages for discovering patients' processes fail at extracting the value and nature of each activity relevant to the whole process. To fill the identified gap, this paper uses a new approach for visualizing patients' pathways as Operational Process Charts (OPC). To do so, Real-Time Location Systems (RTLS) have been used to extract the primary event logs which contain the data related to movements of patients. Next, we have designed a meta-model which can filter and analyze the RTLS event logs and identify the different elements for the construction of OPC's. This paper focuses on presenting the DIAG meta-model for interpretation of event logs and the transformation of these data into opc's.

A hospital is an ecosystem that includes real-time services that require high human interaction on both resources level (doctor, nurses, etc.) and entities level (patients). Designing, planning, improving and controlling this system can be very challenging due to the system complexity governed by several subjective factors that affect the hospital interrelated functions or services. However, continuously changing health care needs that consistently face hospitals require them to keep continuously improving the efficiency of these services as demand increases and as new services are added. This paper proposes a new methodology that uses the concept of Digital Twin (DT) of hospital services based on Discrete Event Simulation (DES) integrated with health care information systems and Internet of things (IoT) devices. It develops a predictive decision support model that employs real-time services data drawn from these systems and devices. This model enables assessing the efficiency of existing health care delivery systems and evaluating the impact of changes in services without disrupting daily activities of the hospital. The developed model, a digital twin (or a virtual replica of the hospital), simulates a number of key hospital health delivery services, based on relevant data retrieved in real-time. Although the model simulates four key services, initially as a proof of concept, but it proposes a general framework, which can be expanded to include other services. The demonstrated proof-of-concept shows that it achieves better planning and improvement of usage of resources, and thus enabling both practitioners and management to examine any model changes to foresee the effectiveness or efficiency of services before they are applied in reality.

Mining useful knowledge from data readily available in today's information systems has been a common challenge in recent years as more and more events are being recorded, and there is need to improve and support many organisational processes in a competitive and rapidly changing environments. The work in this paper shows using a case study of Learning Process - how data from various process domains can be extracted, semantically prepared, and transformed into mining executable formats to support the discovery, monitoring and enhancement of real-time processes. In so doing, it enables the prediction of individual patterns/behaviour through further semantic analysis of the discovered models. Our aim is to extract streams of event logs from a learning execution environment and describe formats that allows for mining and improved process analysis of the captured data. The approach involves augmenting the informative value of the resulting model derived from mining event data about the process by semantically annotating the process elements with concepts they represent in real time using process descriptions languages, and linking them to an ontology specifically designed for representing learning processes to allow for the analysis of the extracted event logs based on concepts rather than the event tags of the process. The semantic analysis allows the meaning of the learning object properties and model to be enhanced through the use of property characteristics and classification of discoverable entities, to generate inference knowledge which are then used to determine useful learning patterns by means of the proposed Semantic Learning Process Mining (SLPM) formalization - described technically as Semantic-Fuzzy Miner. As a result, the approach provides us with the capability to infer new and discover hidden relationships/attributes the process instances share amongst themselves within the knowledge base, and the ability to identify and address the problem of determining the presence of different learning patterns or behaviour. Inference knowledge discovered due to semantic enrichment of the process model is advantageous especially in solving some didactic issues and answering some questions with regards to different Learners behaviour within the context of process mining and semantic model analysis. To this end, we show that information derived from process mining algorithms can be improved by adding semantic knowledge to the resulting model.